Verify against NASA NESC Earth Check Cases

[simnaut]

Goal

Add a verification track that propagates the NASA NESC Earth-orbit check cases through Simulation::step() and compares against the published reference trajectories.

Companion issue to #399 (lunar cases). The NESC GN&C check cases (NESC-RP-12-00767 / "Independent Orbit Propagation Verification") cover Earth-orbit regimes — LEO with non-spherical gravity, drag, SRP, third-body Sun/Moon perturbations, plus GEO and HEO/Molniya cases — with documented ICs, force models, integrator settings, and checkpoint state vectors. They give us an independent cross-check (vendor-neutral, not JEOD/Trick) for the same Earth-orbit regime our existing Tier 3 ISS fleet exercises.

Scope

• Ingest the NESC Earth-case reference data (ICs + checkpoint state vectors) into the astrodyn_verif_nesc crate stood up by Verify against NASA NESC Lunar Check Cases #399, committed under crates/astrodyn_verif_nesc/test_data/earth/.
• Extend the regen binary from Verify against NASA NESC Lunar Check Cases #399 to cover the Earth cases when $NESC_HOME / --nesc-home is supplied.
• Add a tier3_nesc_earth_* test per check case driving astrodyn_runner::Simulation end-to-end from the NESC ICs. Match the case's force model exactly — gravity degree/order, drag model + atmosphere, SRP shadow model, third-body set, integrator, step size — per the No Half-Baked Implementations rule. Tolerances follow the standard 5%-above-observed policy.
• Wire parity-trait siblings so the Bevy adapter inherits Earth-case coverage transitively (issue Make bevy_parity tests a superset of every runner↔JEOD Tier 3 scenario #389 infrastructure), with KNOWN_PARITY_GAPS exemptions only for cases that genuinely can't ride the recipe path.

Out of scope

• The lunar / cislunar cases — those live in Verify against NASA NESC Lunar Check Cases #399. Keep the splits clean so a regression in one regime doesn't mask the other.
• Any new physics. If a case requires a force model we don't yet have (e.g., a specific drag coefficient model, a third-body configuration we haven't ported), that's a separate port issue, not a tolerance adjustment.

Open questions

• The NESC Earth suite spans several gravity-degree variants (point mass, J2-only, full harmonics). Which subset is "canonical" for an initial pass — start with point mass / J2 cases, or go straight to the full harmonics cases since we already have GGM05C wired?
• Atmosphere: NESC cases typically specify a Jacchia or NRLMSISE model. We currently have MET. Does the case allow a model-equivalent substitution, or do we need to port the specified model first?
• Overlap with our existing JEOD ISS Tier 3 tests: should the NESC LEO cases run alongside, or replace, any of the existing tier3 fixtures?